Update to Superpowers

{
  "name": "executing-plans",
  "description": "Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 129
    },
    {
      "relative_path": "scripts/task-done",
      "size_in_bytes": 1670
    },
    {
      "relative_path": "scripts/task-start",
      "size_in_bytes": 963
    }
  ],
  "skill_md_contents": "---\nname: executing-plans\ndescription: Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available\n---\n\n# Executing Plans\n\nExecute the plan yourself, task by task, in this session: no implementer\nsubagent per task, no reviewer per task. One fresh-context review of the\nwhole branch at the end.\n\n**Why inline:** Subagent-driven development pays for a fresh implementer\nand a fresh reviewer on every task, each re-reading the codebase from zero.\nInline execution pays for one context (yours) plus one reviewer at the end.\nWhat it gives up is a fresh context per task and a second pair of eyes per\ntask. This skill keeps what those two things bought, by other means: the\nbrief is the spec, the ledger is your memory, TDD is the per-task gate, and\nthe final reviewer is the second pair of eyes.\n\n**Core principle:** The plan already did the thinking. Execute it exactly,\nprove each step with a test you watched fail and then pass, and leave a\nrecord that survives your own forgetting.\n\n**Narration:** between tool calls, narrate at most one short line — the\nledger and the tool results carry the record.\n\n**Continuous execution:** Do not pause to check in with your human partner\nbetween tasks. They chose inline execution to spend less, not to answer\n\"should I continue?\" after every task. Execute all tasks from the plan\nwithout stopping.\n\n**Rulings, not stalls.** Conflicts, ambiguities, plan defects — decide them.\nThe spec is the binding authority, the plan is its argument, and your\njudgment settles what neither answers. Record every decision in the ledger\nas `Ruling: <what you decided> — <why> — <what it costs if wrong>`, and keep\ngoing. Deviating from the plan without a ledgered ruling is a decision made\nin secret.\n\nFour things stop you, and only these: an irreversible or destructive\noperation; a security-sensitive action; a side effect outside this worktree\nthat norms say you ask about first (a merge, a push to a shared branch, a\npublish); and a plan so broken that every path forward is a guess. For\nthose, stop and ask.\n\n## When to Use\n\n- You have a plan from superpowers:writing-plans and your human partner\n  chose inline execution at the handoff.\n- Your harness has no subagent tool (see the per-platform references in\n  `../using-superpowers/references/`). Never fabricate a dispatch; run\n  the plan here.\n- Tasks are mostly independent — the same precondition as\n  superpowers:subagent-driven-development.\n\nA fully specified plan makes inline execution transcription plus testing:\nit runs well on a mid-tier session model, and the one place the most\ncapable model earns its cost is the final review, which this skill\ndispatches separately. Tell your human partner so when they choose inline.\n\nPrefer superpowers:subagent-driven-development when your human partner\nwants a review gate on every task, or when the plan is long enough that\nits later tasks would run on a compacted context. Inline execution over a\nlong plan still works — the ledger is what makes it recoverable — but the\nlast tasks get the least of you.\n\n## The Process\n\n```dot\ndigraph process {\n    rankdir=TB;\n\n    subgraph cluster_per_task {\n        label=\"Per Task\";\n        \"task-start: brief + BASE; read the brief\" [shape=box];\n        \"Work the steps in order: TDD, run every verification, read every output\" [shape=box];\n        \"Step output matches plan's Expected?\" [shape=diamond];\n        \"Plan wrong? Rule and ledger. Code wrong? systematic-debugging\" [shape=box];\n        \"Commit as the plan's commit steps say\" [shape=box];\n        \"Completion contract met?\" [shape=diamond];\n        \"task-done: run tests, ledger the result; mark todo complete\" [shape=box];\n    }\n\n    \"Setup: worktree, workspace + ledger, read plan + spec, pre-flight scan\" [shape=box];\n    \"More tasks remain?\" [shape=diamond];\n    \"Final whole-branch review (fresh reviewer if you have one)\" [shape=box];\n    \"Re-grade, then: Critical/Important → ONE fix pass, each fix RED→GREEN + green suite; Minor → ledger\" [shape=box];\n    \"Final review clean: delete this plan's workspace\" [shape=box];\n    \"Use superpowers:finishing-a-development-branch\" [shape=box style=filled fillcolor=lightgreen];\n\n    \"Setup: worktree, workspace + ledger, read plan + spec, pre-flight scan\" -> \"task-start: brief + BASE; read the brief\";\n    \"task-start: brief + BASE; read the brief\" -> \"Work the steps in order: TDD, run every verification, read every output\";\n    \"Work the steps in order: TDD, run every verification, read every output\" -> \"Step output matches plan's Expected?\";\n    \"Step output matches plan's Expected?\" -> \"Plan wrong? Rule and ledger. Code wrong? systematic-debugging\" [label=\"no\"];\n    \"Plan wrong? Rule and ledger. Code wrong? systematic-debugging\" -> \"Work the steps in order: TDD, run every verification, read every output\";\n    \"Step output matches plan's Expected?\" -> \"Commit as the plan's commit steps say\" [label=\"yes, last step\"];\n    \"Commit as the plan's commit steps say\" -> \"Completion contract met?\";\n    \"Completion contract met?\" -> \"Work the steps in order: TDD, run every verification, read every output\" [label=\"no - finish the task\"];\n    \"Completion contract met?\" -> \"task-done: run tests, ledger the result; mark todo complete\" [label=\"yes\"];\n    \"task-done: run tests, ledger the result; mark todo complete\" -> \"More tasks remain?\";\n    \"More tasks remain?\" -> \"task-start: brief + BASE; read the brief\" [label=\"yes\"];\n    \"More tasks remain?\" -> \"Final whole-branch review (fresh reviewer if you have one)\" [label=\"no\"];\n    \"Final whole-branch review (fresh reviewer if you have one)\" -> \"Re-grade, then: Critical/Important → ONE fix pass, each fix RED→GREEN + green suite; Minor → ledger\";\n    \"Re-grade, then: Critical/Important → ONE fix pass, each fix RED→GREEN + green suite; Minor → ledger\" -> \"Final review clean: delete this plan's workspace\";\n    \"Final review clean: delete this plan's workspace\" -> \"Use superpowers:finishing-a-development-branch\";\n}\n```\n\n## Setup\n\nEnsure the work happens in an isolated workspace: use\nsuperpowers:using-git-worktrees to create one or verify the existing one.\nNever start implementation on a main/master branch without your human\npartner's explicit consent.\n\nConversation memory does not survive compaction. An inline executor that\nloses its place re-implements tasks whose commits already exist — the same\nfailure as a controller re-dispatching them, paid for in your own context.\nTrack progress in a ledger file, not only in todos. Harness todos are a\nlive view; the ledger is the record.\n\nThe workspace and ledger are shared with superpowers:subagent-driven-development\n— same directory, same format — so a plan can change executors mid-flight\nand the new one resumes from the same ledger.\n\n- Each plan owns a workspace: at skill start, run\n  `../subagent-driven-development/scripts/sdd-workspace PLAN_FILE` — it\n  prints the plan's git-ignored directory\n  (`<repo-root>/.superpowers/sdd/<plan-basename>/`), home to every\n  artifact for THIS plan: ledger, briefs, review packages. Another plan's\n  directory is never yours to read or write.\n- Check for this plan's ledger at `<workspace>/progress.md`. If its first\n  line names your plan file, tasks with a `Task <N>: complete` line are\n  DONE — do not redo them; resume at the first task without one. Their\n  commits exist in git even when your context no longer remembers making\n  them: after compaction, trust the ledger and `git log` over your own\n  recollection. A ledger whose first line names a different plan file is\n  another plan's progress: leave it and start your own, fresh.\n- Create the ledger with its identity as the first line:\n  `# SDD ledger — plan: <plan file path>`.\n- `git clean -fdx` will destroy the workspace (it's git-ignored scratch);\n  if that happens, recover from `git log`.\n\nRead the plan once, note its context and Global Constraints, and create a\ntodo per task. If the plan names a Spec, read that too: the spec is the\nauthority the plan argues from, and conflicts inside the plan resolve\nagainst it. A plan with no reachable spec gets a ledger note saying so —\nrulings made without one are provisional.\n\n**REQUIRED SUB-SKILL:** load superpowers:test-driven-development now,\nbefore Task 1. It governs every step of every task below; a plan whose\nsteps already say \"write the failing test first\" does not exempt you\nfrom reading it.\n\nBefore Task 1, scan the plan for conflicts between tasks. The plan's\nInterfaces blocks tell you where to look: for every task that consumes\nwhat an earlier task produces, one ledger row — the two tasks, what one\nproduces against what the other consumes, and what you found. Tasks that\nshare nothing get no row; a plan whose tasks share nothing gets the single\nline `Pre-flight: no shared interfaces`. Rule on each conflict a row\nsurfaces with the spec as the binding authority, record the ruling beside\nits row, and start Task 1. Each task's own text is checked when you read\nits brief, not here.\n\n## The Task Loop\n\nEverything you print, and every tool result, stays resident in your\ncontext for the rest of the session. Redirect long test output to a file\nin the workspace and read its tail; read a brief, not the whole plan.\n\n### 1. Take the task\n\n- Run this skill's `scripts/task-start PLAN_FILE N`. It prints the brief\n  path and BASE (the commit the task's range is cut from) in one call.\n  Read the brief for every task, including ones you remember from setup:\n  what you remember is a summary, the brief has the exact values,\n  signatures, and test cases.\n- Mark the task's todo in_progress.\n\nEvery tool call is a turn that re-reads your whole context. Bookkeeping\nrides along with work — a ledger append in the same call as the commit,\nnever in a call of its own.\n\n### 2. Work the steps\n\nThe plan's steps are already in RED-GREEN order; follow them in that\norder under superpowers:test-driven-development, loaded at setup. A test\nstep's code is written first and run first. Watching it fail is a step,\nnot a formality — a test that passes before the implementation exists is\na finding about the test.\n\nEvery step that runs a command has an `Expected:` line. Run the command,\nread its output, and compare. Three outcomes:\n\n- **Matches.** Next step.\n- **The code is wrong.** Use superpowers:systematic-debugging. Find the\n  cause; never patch the symptom to make the step's output match.\n- **The plan is wrong** — a step contradicts the spec, an interface from an\n  earlier task doesn't match what this task consumes, a command that\n  cannot work. Rule on the smallest change that satisfies the spec, ledger\n  it as `Task <N>: Ruling: <finding> — <what you decided and why>`, and\n  continue. The ruling is carried, not remembered: later tasks that touch\n  the same interface read it from the ledger.\n\nCommit as the plan's commit steps say. A task that spans several commits\nis fine; BASE is what the review range is cut from, never `HEAD~1`.\n\n### 3. The completion contract\n\nBefore a task's ledger line, all of the following are true, with evidence\nin this session — not inferred from the diff looking right:\n\n- Every test the brief names exists and ran in this task, and you read\n  the output.\n- The final test run for the task passed — `task-done` is that run, and\n  it writes the command and result into the ledger line.\n- Every `Expected:` line in the brief was compared against real output.\n- Every deviation from the brief has a `Ruling:` line in the ledger.\n\n**REQUIRED SUB-SKILL:** superpowers:verification-before-completion governs\nthe claim. If any item is missing, the task is not complete: finish it.\n\n### 4. Complete the task\n\nRun this skill's `scripts/task-done PLAN_FILE N BASE -- <test command>`\nwith the test command the brief names for the whole task. It runs the\ntests, keeps the full output in the workspace, prints the tail, and — only\nif they pass — appends the completion line to the ledger:\n\n`Task <N>: complete (commits <base7>..<head7>, tests: <command> → <result>)`\n\nA failing run records nothing; the task is not complete. When it records,\nmark the todo complete and take the next task.\n\n## Final Review\n\nRun `../subagent-driven-development/scripts/review-package PLAN_FILE MERGE_BASE HEAD`\n(MERGE_BASE = the commit the branch started from, e.g.\n`git merge-base main HEAD`) and review from the file it prints.\n\n**With a subagent tool:** dispatch the reviewer on the most capable\navailable model — the whole-branch review is a judgment task — using\nsuperpowers:requesting-code-review's\n[code-reviewer.md](../requesting-code-review/code-reviewer.md), with the\npackage path, the plan and spec paths, the plan's Review Focus section\nverbatim if it has one (the input classes and failure modes the plan's\ntests do not exercise — the reviewer checks each deliberately), and a\npointer to the ledger's `Ruling:` lines so it can weigh the calls you\nmade. Specify the model\nexplicitly; an omitted model inherits the session's, which may not be the\nmost capable. This is the one fresh context the whole run buys. Do not\nskip it, and do not replace it with your own read of the diff.\n\n**Without a subagent tool:** read code-reviewer.md and perform that review\nyourself against the package, as a separate pass after the last task's\nledger line. Write `Final review: self-review (no subagent tool)` to the\nledger, and say so in your final message: a self-review by the author is\nweaker than a fresh reviewer, and your human partner decides whether that\nis enough before merge.\n\nSort the findings before you act on any of them. The reviewer's severity\nlabels are advice; the gate is yours. Its \"Declined to judge\" list is\nyours too: every line there is a ruling you make and ledger, exactly like\na plan conflict — `Final: Ruling: <behavior the reviewer set aside> —\n<what a reasonable person using this software gets, and why that stands\nor why it is now a finding> — <cost if wrong>`. Re-grade first, by effect: the\nspec is a vision document, and a finding's grade is what a reasonable\nperson using this software gets if it ships, not whether the spec names\nthe input that triggers it — a reviewer who set a finding at Minor\nbecause the spec was silent has graded the spec, not the effect. Then:\n\n- **Critical and Important** enter the fix pass.\n- **Minor** goes to the ledger as `Final: minor (deferred): <one-liner>`\n  and to your final message under \"Deferred minors\". Minors never enter\n  the fix pass, and never become rulings — a ruling is a decision about a\n  conflict, not a note that you declined a polish suggestion.\n\nFix the Critical and Important findings yourself — you are the\nimplementer here — in ONE pass. Each fix is verified by TDD, not by a\nsecond reviewer: write the test that reproduces the finding, watch it\nfail, make it pass, then run the whole suite. Record each in the ledger as\n`Final: fixed <finding> — <test name> RED→GREEN, suite <N>/<N>`. A fix\nwithout a test that failed first is not verified; a suite that is not\ngreen after the pass means the pass is not over. Do not dispatch a\nre-review: it would re-read a diff whose covering tests already answer\n\"addressed\" and whose suite run already answers \"broke nothing\".\n\nA finding you decide not to fix is a ruling — `Final: Ruling: <finding> —\n<why the code stands> — <cost if wrong>` — and reaches your human partner\nin the rulings list. There is no second fix pass.\n\n## Finish\n\nBefore you delete anything, collect every ledger line containing\n`Ruling:` into your final message under \"Rulings I made\", in the order you\nmade them, each with what it costs if wrong, and every `minor (deferred)`\nline under \"Deferred minors\". Both lists are exhaustive. Your final\nmessage is the only place the decisions you took on your human partner's\nbehalf — and the findings you chose not to act on — reach them.\n\nWhen the final review is clean and its fixes are committed, delete this\nplan's workspace directory — the git history is the record now. Sibling\ndirectories belong to other plans; leave them alone.\n\nUse superpowers:finishing-a-development-branch.\n\n## Common Rationalizations\n\n| Excuse | Reality |\n|--------|---------|\n| \"I remember what Task N says\" | You remember a summary. The brief has the exact values. Read it. |\n| \"The plan's code is right, skip watching the test fail\" | A test you never saw fail proves nothing. It is one step. Run it. |\n| \"I'll run the full suite at the end instead of per step\" | Per-step runs are how you learn which step broke it. The end-of-task run is the contract, not a substitute. |\n| \"The plan is wrong here, I'll just do the right thing\" | Do the right thing and ledger the ruling. Unledgered deviation is a decision made in secret. |\n| \"I'll write the ledger lines after a few tasks\" | Compaction does not wait for a convenient moment. One line per task, in the same message as the commit. |\n| \"Let me check in before the next task\" | They chose inline to spend less. Progress prompts spend their time instead. Only the four stops stop you. |\n| \"I read my own diff carefully; the final reviewer is redundant\" | Same author, same blind spots. The reviewer is the only fresh context this run buys. |\n| \"Tests should pass, the change was trivial\" | \"Should\" is not evidence. The contract requires the command and its output. |\n| \"Subagents are slow and expensive, I'll skip the final review too\" | Inline already removed the per-task reviewers. One review of the whole branch is the floor, not the ceiling. |\n| \"The reviewer said Minor, so it's Minor\" | The label graded the spec's silence. Grade what the person gets. Re-grade, then gate. |\n| \"The fix is obvious, no need for a failing test first\" | The failing test is the only proof the finding was real and is now gone. Without it you have a diff and a hope. |\n| \"I'll fix the minors too while I'm in there\" | Every minor you fix is a test, a fix, and a suite run your partner did not ask for. Ledger them; your partner decides. |\n\n## Example Workflow\n\n```\nYou: I'm using the executing-plans skill to implement this plan inline.\n\n[Setup: worktree verified]\n[Read plan once: docs/superpowers/plans/feature-plan.md; spec read]\n[Resolve workspace: sdd-workspace docs/superpowers/plans/feature-plan.md — no ledger inside, fresh start]\n[Pre-flight scan: 2 shared-interface rows, 4 self-consistency rows, clean; written to ledger]\n[Create todos for all tasks]\n\nTask 1: Hook installation script\n\n[task-start plan 1 → brief read; BASE a1b2c3d]\n[Step 1: write failing test — written]\n[Step 2: run it — FAIL: install_hook not defined. Matches Expected.]\n[Step 3: implement — written]\n[Step 4: run it — PASS 1/1. Matches Expected.]\n[Step 5: commit — d4e5f6a]\n[Contract: tests ran, output read, no deviations]\n[task-done plan 1 a1b2c3d -- npm test -- hooks → ledger: Task 1: complete (commits a1b2c3d..d4e5f6a, tests: npm test -- hooks → 1/1 pass)]\n\nTask 2: Recovery modes\n\n[task-start plan 2 → brief read; BASE d4e5f6a]\n[Step 2: run failing test — FAIL, but on an import error: Task 1 exported\n installHook, brief consumes install_hook]\n[Ruling: brief's consumer name is a typo against Task 1's Produces block;\n use installHook — Ledger: Task 2: Ruling: install_hook → installHook — matches Task 1 Produces — cost if wrong: one rename]\n[Steps 2-5 as planned; commit b7c8d9e]\n[task-done plan 2 d4e5f6a -- npm test -- recovery → ledger: Task 2: complete (commits d4e5f6a..b7c8d9e, tests: npm test -- recovery → 8/8 pass)]\n\n...\n\n[After all tasks: review-package plan MERGE_BASE HEAD; dispatch code-reviewer, most capable model]\nReviewer: One Important finding — progress reporting interval hardcoded. Two Minor.\n[Re-grade: Important stands; minors → ledger as deferred]\n[Fix pass: test_progress_interval_configurable RED → extract PROGRESS_INTERVAL → GREEN; suite 12/12; commit]\n[Ledger: Final: fixed hardcoded interval — test_progress_interval_configurable RED→GREEN, suite 12/12]\n\nRulings I made:\n- Task 2: install_hook → installHook (brief typo; cost if wrong: one rename)\n\nDeferred minors:\n- README lacks a usage example\n- recovery.js could split verify/repair into two files\n\n[Delete this plan's workspace — the record now lives in git]\n\nUsing superpowers:finishing-a-development-branch.\n```\n"
}